Abstract
In this article, we address the super-resolution problems, which estimate the high-resolution multispectral images from the multispectral Sentinel-2 (S2) images with different resolutions. Since S2 images can be naturally represented by tensors, we reformulate the degradation process as the tensor-based form. Based on the degradation mechanism, we build a tensor-based optimization model for S2 images super-resolution problem, which fully exploits intrinsic nonlocal spatial similarity and global spectral redundancy. Specifically, the model consists of the data fidelity term and the low-multirank regularizer tailored to thoroughly mining the inherent spatial-nonlocal and spectral redundancy. Then, we develop an efficient alternating direction method of multipliers algorithm with theoretically guaranteed convergence to tackle the resulting tensor optimization problem. Numerical experiments including simulated and real data demonstrate that our method outperforms the competing methods visually and qualitatively.
Highlights
I N REMOTE sensing, an increasing number of satellites are launched to acquire multispectral (MS) images, which are used to perform terrestrial observations in support of services such as environmental monitoring, land cover changes detection, and natural disaster management [1]–[3]
1) We suggest a S2 super resolution (SR) model whose data fidelity term depicts the tensor-based degradation process and the regularization term fully exploits the nonlocal spatial similarity and global spectral redundancy of S2 images by using a logdet-based nonconvex surrogate of tensor multirank regularizer
We evaluate the performance of the proposed algorithm on simulated data with S2 parameters and real S2 images
Summary
I N REMOTE sensing, an increasing number of satellites are launched to acquire multispectral (MS) images, which are used to perform terrestrial observations in support of services such as environmental monitoring, land cover changes detection, and natural disaster management [1]–[3]. Due to the restrictions of imaging gadgets, there is a tradeoff among spatial and spectral resolutions of MS images, i.e., the spatial resolution [or ground sampling distance (GSD)] of images acquired by sensors varies according to different spectral bands. To obtain a higher signal-to-noise ratio (SNR), the spatial resolution has to be lower if the spectral one is required to be higher, so optical images might be blurry. Losing spectral resolution is the price to pay for a high spatial resolution.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have